This paper provides a metric for determining whether a given pair of English words is perceived to be morphologically related, based on objective measurements of the words’ orthographic, phonetic, and semantic similarity to each other. The metric is developed on the basis of results from a behavioural study in which participants were asked to judge the relative similarity of pairs of words. The metric is intended to help researchers determine which forms in a language plausibly have segments that alternate; as an example, it is applied to the lexicon of English to illustrate its utility in calculating the frequency of alternation of [s] and [ʃ].
